Redox signaling affects the heart both physiologically (e.g., cell growth, differentiation, excitation-contraction coupling) and pathologically (e.g., fibrosis, cardiac remodeling, hypertrophy) (Burgoyne et al. 2012, Santos et al. 2011). Cells are equipped with redox-stress defense mechanisms that include endogenous antioxidants (e.g., glutathione, glutathione peroxidase, thioredoxin, peroxiredoxin) (Hafstad et al. 2013, Santos et al. 2011). Therapeutic application of compounds with antioxidant activity has been proposed as an approach to protect the heart, with however so far limited success. A key challenge to the field is to determine the mechanisms underlying the fine-tuning of the redox status in cells and its subcellular compartments. The application of redox assays/sensors promises for detailed insight into qualitative and quantitative changes of the redox state of the cell. Until recently, most studies have utilized a variety of redox sensitive dyes, such as dichlorodihydrofluorescein diacetate, amplex red, and dihydrorhodamine (Forkink et al. 2010, Griendling et al. 2016, Meyer and Dick 2010) to monitor both ROS in and redox states of cells. The recent advent of genetically encoded redox sensors promises to advance the field markedly because of the possibility to target sensors to subcellular compartments (Swain et al. 2016) or cell types, as demonstrated in this thesis.
The following main observation could be made:
1) Human fibroblasts and cardiomyocytes can be stably transduced with the cytosolic glutathione redox sensor Grx1-roGFP2 without signs for transgene toxicity.
2) Cardiomyocytes are more sensitive to oxidation and less reactive to reduction compared to fibroblasts.
70 3) This may at least in part be explained by the more reduced glutathione redox
potential in cardiomyocytes compared to fibroblasts.
4) Fibroblasts as compared to cardiomyocytes displayed a wider range in redox state alterations.
5) Fibroblasts show faster responses to oxidative and reductive challenges compared to cardiomyocytes.
6) Assessment of redox alterations under pharmacological stimulation was confounded by oxidizing effects of DMSO.
7) Glutathione redox sensing can be performed in a cell type specific manner in EHM and thus may allow dissecting the “redox-interplay” between different cell types in a tissue context.
Fibroblast from human foreskin and cardiomyocytes derived from the human embryonic stem cell line HES2 were utilized as surrogates for the most abundant cell types of the heart. The suitability of these cells to model human heart function was recently demonstrated (Tiburcy et al. 2017).
4.1 Tools for redox quantification in living cells
Monitoring of redox alteration in cells and tissue requires tools for the robust and quantitative reporting of the dynamic changes in cellular redox states. The most promising tools to monitor cellular redox changes dynamically, in real time and compartment specific appear to be biological sensors genetically integrated into cells of interest (Lukyanov and Belousov 2014, Meyer and Dick 2010). In this study, we utilized the genetically encoded glutathione redox sensor Grx1-roGFP2, which detects redox changes of the GSH:GSSG couple in a quantitative manner in real time. Modifications of the Grx1-roGFP2 sensor have been shown to also allow for subcellular targeting of the redox sensor (Swain et al. 2016). This study identified differences in the cytosol and mitochondria of murine cardiomyocytes. In this thesis, targeting of the most abundant cell types in the human heart was demonstrated by
71 transducing the cytosolic Grx1-roGFP2 sensor into cardiomyocytes and fibroblasts.
Human embryonic stem (HES) cell-derived cardiomyocytes and human foreskin fibroblasts were chosen because of their accessibility and the finding that these cell types can be reconstituted to myocardium with structural, molecular, and functional properties of the postnatal human heart (Tiburcy et al. 2017). In fact, human foreskin and adult heart derived fibroblasts are similarly supportive to heart muscle reconstitution in vitro (unpublished data). Lentiviral transduction was chosen in a first attempt to validate the concept of Grx1-roGFP2 expression in human cardiomyocytes and fibroblasts cells. The need for repeated transduction especially in HES-derived cells makes this a tedious approach, which will have to be replaced by the stable introduction of the reporters by homologous recombination. In addition, silencing and incomplete labelling of the cells of interest is an issue that can be circumvented with for example TALEN or CRISPR mediated integration of target sequences in defined genomic loci such as the AAVS1 site (Mussolino et al. 2014). Despite these caveats it was possible to establish proof-of-concept for glutathione redox potential sensing in human heart cells. Validation experiments with oxidizing (H2O2, DA) and reducing (DTT) compounds confirmed the principle utility of the Grx1-roGFP2 sensor in human cardiomyocytes and fibroblasts. Some of the observed variability may be contributed to variable transduction and activity of the Grx1-roGFP2 sensor.
Another robust way to analyze cellular oxidation is by the investigation of characteristic protein modifications under oxidative stimulation. For example, effective oxidation of peroxiredoxin by H2O2 should lead to its sulfonylation (-SO3) (Sobotta et al. 2013), which can be detected by Prx-SO3 specific antibodies.
Immunoblots confirmed that H2O2 at the chosen concentration range would indeed result in oxidation of Prx with an apparent H2O2 concentration dependent effect at 10 - 1,000 µmol/L H2O2.
4.2 Cell type specific glutathione redox responses
Both HFFs and HES2-CMs demonstrated a concentration dependent oxidizing and reducing trajectory with distinct sensitivities (EC50), response kinetics (t50), and EGSH
(Table 11).
72
Table 11: Summary of Grx1-roGFP2 sensor activites in HES2-CM and HFF.
n = numbers of cells analysed.
The more reduced EGSH in cardiomyocytes was a particularly interesting observation and has to be considered when interpreting the observed differences in the responses to oxidative and reductive challenges. The response range to oxidative and reductive stimulation with H2O2 and DTT, respectively, was for both cell types 10 - 1,000 µmol/L and 0.01 – 1 mmol/L. HES2-CM showed a markedly higher sensitivity to oxidation (refer to EC50 in Table 11) with at the same time a clearly smaller dynamic range in redox alterations (refer to R/R0 in Table 11); cardiomyocytes
73 demonstrated a ΔR/R0 differences of ~80% vs. ~140% in fibroblasts (absolute % change). Another interesting observation was the apparent roGFP-oxidation reserve in fibroblasts, which was evidenced by the lower R/R0 signal change under maximal H2O2 (~80%) vs. DA (~150%).
Fibroblasts appear to exhibit a particular high expression in NOX4 (RNA-sequencing data in (Tiburcy et al. 2017)). NOX4 is constitutively active and H2O2 producing (Kuroda et al. 2010) and moreover directly activated by H2O2 mediated oxidation (Colston et al. 2005). This suggests a more extensive oxidant load in fibroblasts and thus may on the one hand explain the difference in EGSH (Table 11) and the wider
reported in this thesis requires more detailed investigations of differences in transmembranous H2O2 gradients as well as subcellular targets in cardiomyocytes and fibroblasts. The recent demonstration of cytosol and mitochondria targeting of the Grx1-roGFP2 sensor is particularly interesting in this context (Swain et al. 2016).
4.3 Cell type specific redox responses to angiotensin II and drugs
Another aspect of this study was to investigate cell type redox responses upon stimulation with angiotensin II and inotropic drugs. From previous studies, stimulation with angiotensin II, the key effector of rennin angiotensin system, is known to be involved in ROS production pathways by stimulating membrane bound NAD(P)H oxidase (Dikalov and Nazarewicz 2013, Griendling et al. 1994, Vazquez-Medina et al.
2013). In our experiments, HFFs did not exhibit obvious redox changes over the whole range of angiotensin II concentrations tested; surprisingly, a reductive effect was observed under 100 nmol/L angiotensin II. This is in contrast to the study conducted by Sano et al. and Swain et al. They demonstrated ROS production and roGFP2 oxidation in cardiac fibroblasts by angiotensin II stimulation (Sano et al.
74 2001, Swain et al. 2016). HES2-CMs showed however some oxidative effects at a low angiotensin concentration (0.1 nmol/L). These results are surprising and somewhat inconsistent. Stable expression of Grx1-roGFP2 in every cell of interest with a similar expression level may help to further clarify these apparently discrepant findings. Collectively, these data suggest that assessments of oxidation and reduction would benefit from cell-type specific analyses.
In an attempt to study oxidative and reductive drug effects, levosimendan and omecamtiv mecarbil were applied to HES2-CM and HFF. These cardioactive inotropes are applied in acute heart failure with cardioprotective effects reported for levosimendan (via activation of the mitochondrial ATP-sensitive K+-channels) and no effect on myocardial oxygen consumption reported for omecamtiv mecarbil (Farmakis et al. 2016, Meijs et al. 2012). There was no obvious effect of LEVO and OME in cardiomyocyte or fibroblasts oxidation. However, these experiments appeared to be severely confounded oxidizing effects of the solvent DMSO. A refinement of the study protocol (avoidance of DMSO) is needed to ensure the assessment of drug specific effects and confirm the present findings of no effects of LEVO and OME on the redox state of cardiomyocytes and fibroblasts.
4.4 Redox sensing in a heart muscle tissue context
Up to this point, we discussed results and analysis that were performed in monolayer cardiomyocytes and fibroblasts. Cellular measurements provide information regarding compound effects and pathways involved in redox changes in specific cell cultures. However, investigations in a cardiac muscle model that mimic the microenvironment of the heart tissue are physiologically more relevant than monolayer cell culture. Furthermore, in a tissue context the interaction between cells plays a key role in the investigation of drug effects on redox homeostasis in a cardiac tissue. EHMs exhibit a higher degree of maturity as compared to monolayer cultures (resemble an embryonic phenotype); in fact, they reflect in many structural, molecular and functional parameters properties of the postnatal human heart (Tiburcy et al.
2017). By genetically modifying the two most abundant heart cell types and mixing them in a defined context, it became possible to define response to redox challenges in a heart muscle context. Importantly, we demonstrated that the biosensor
75 responded in EHM to the bolus of H2O2 and DTT both optically and functionally.
Stimulation of human EHMs with oxidizing and reducing agents confirmed that we were able to detect redox changes occurring within the tissue. Interestingly, contractile properties were deteriorated under maximal oxidation and appeared to be enhanced under maximal reduction with H2O2 and DTT, respectively. These results fit well with previous data showing that redox alterations affect signaling pathways (eg. Ca2+/calmodulin-dependent kinase II; CAMKII or cAMP-dependent protein kinase; PKA) important in contractility and could bring cardiomyocyte to death (Santos et al. 2016).
These studies will be extended in the future to study the interplay between the different cardiac cell species and their specific responses to environmental, biomechanical, and pharmacological stimuli. In addition, a link to signaling pathways will need to be established and assessments of acute vs chronic effects will need to be performed. A potential caveat is the thickness of engineered tissue (~ 1 mm) and the intra-tissue distance between cells (from direct contact to several 100 µm), which may limit cell-cell communication via highly reactive and thus instable oxidants.
However, effects of oxidation (intracellular or extracellular) of secreted proteins, such as extracellular matrix proteins or growth factors, could be studied and may provide relevant information on redox-mediated tissue homeostasis. Alternative technologies for intra-tissue assessment of oxidation were recently developed (Fujikawa et al.
2016) and rely on the arrest of the redox state of the roGFP-biosensor by the membrane-permeable thiol-alkylating agent (NEM). First experiments in human EHMs suggest that this method would indeed be applicable to study the redox state histologically. This together with the possibility for in tissue fluorescence analyses by confocal (~50 µm penetration depth) and 2-photon (~200 µm) microscopy should allow for comprehensive in tissue phenotyping of redox mechanisms and their association with heart muscle function.
76